Re:Leaked Apple M5 9 core Geekbench scores
by rowanG077 on Tuesday September 30, 2025 @09:04PM
(#45433188)
No. I am under the impression that there is no replacement on the horizon that could take on the challenge of further optimizing the GPU driver and work on M3/M4/M5 GPU support once upstreaming efforts have settled down.
Re:Leaked Apple M5 9 core Geekbench scores
by GeekyBear on Tuesday September 30, 2025 @09:22PM
(#45433294)
Alyssa wasn't doing the GPU driver work.
Marcan already provided the tools needed to capture the data being passed back and forth between MacOS and the GPU, so you can see exactly what the newer versions of the SOC are doing that is different.
Re:Leaked Apple M5 9 core Geekbench scores
by rowanG077 on Tuesday September 30, 2025 @09:37PM
(#45433402)
I think you are confused. There are two drivers here. The kernel space driver mostly developed by lina. And the user space driver developed by Alyssa. Both are drivers. And of the two the user space driver is an order of magnitude more complex.
Re:Leaked Apple M5 9 core Geekbench scores
by GeekyBear on Tuesday September 30, 2025 @10:09PM
(#45433593)
Alyssa's userspace work that I've already mentioned has been completed and upstreamed?
> We are pleased to announce that our graphics driver userspace API (uAPI) has been merged into the Linux kernel. This major milestone allows us to finally enable OpenGL, OpenCL and Vulkan support for Apple Silicon in upstream Mesa.
Re:Leaked Apple M5 9 core Geekbench scores
by rowanG077 on Tuesday September 30, 2025 @10:44PM
(#45433789)
Sure for M1/M2 it's somewhat correct that it is "completed". M3 and beyond have a lot of GPU differences though(A completely new microcode afaik), so the loss is still traumatic. I also don't think GPU perf is on the level of OSX though I did not do a recent one to one bench.
Are you under the impression that Linux is about to abandon OpenGL and Vulkan in favor of a new graphics API that only Alyssa could possibly implement?